>>21
That's pretty cool. It seems a fairly efficient tiling to calculate, but very low quality. There is a lot of high-level repetition in it even if the tiling itself is aperiodic. For example every other tile in every other row is blue. Every fourth tile in every fourth row is purple. You will see these artifacts in the final product.
Check out what this guy has done:
http://www.dgp.toronto.edu/people/stam/reality/Research/pdf/R046.pdf
This tiling is aperiodic and exhibits very little visible high-level structure. I would guess that it's because the tiling there is based on a substitution system rather than being constructive like yours, but I have nothing to substantiate that claim. Nevertheless the screenshots look good.
Have you written a renderer yet? Ideally you'd write a tile-based renderer that takes an arbitrary tiling and lets you pan/scroll with mouse+wheel. You should be able to write this with SDL/GLFW in a couple hundred lines of C. Then test out the different algorithms.